[AMBER] How does tleap detect improper torsions?

From: Brian Radak <brian.radak.gmail.com>
Date: Tue, 22 Sep 2020 16:57:46 -0400

Hopefully there's a tleap guru who has some ideas about the following:

We have a work flow that generates frcmod files for bespoke
parameterization of small organic molecules. On /some/ occasions tleap
gives the very clear warning

** Warning: No sp2 improper torsion term for aa-bb-cc-dd

Where I've used dummy types for illustrative purposes. What's confusing
is that the loaded frcmod clearly has an improper defined for
aa-cc-bb-dd -- why does tleap ignore this? When I swapped the atom order
manually, the warning goes away and I've confirmed that one extra
improper appears in the prmtop.

 From poking around the archives I understand that multiple equivalent
improper terms are a big no-no, is that related? Our procedure (which I
didn't develop) is very careful to enforce that the third atom is the
central atom all cases - is that what tleap searches for? If so, how?
Our code does this by:

1.) Checking at all atoms in the improper torsion are bonded to atom3.
If yes - yay, this torsion is good
2.) If no, check all possible connectivity patterns until one atom is
bound to all others
3.) Swap the new central atom with atom3 in the definition

Is there maybe a sneaky bug in this algorithm? So far, I've only noticed
this problem with torsions that include all heavy atoms, but that might
just be coincidence.

Thanks in advance,

Brian Radak, PhD
Senior Investigator, Silicon Therapeutics
AMBER mailing list
Received on Tue Sep 22 2020 - 14:00:02 PDT
Custom Search